Description of Products
The new gridded GFS MOS QPF products are developed with standard MDL MOS procedures, where the NCEP GFS is the driving NWP model. The QPFs estimate 6-h precipitation amounts in HRAP 4-km grid boxes, as specified in quality-controlled RFC Stage III precipitation analyses. The geographical coverage spans the CONUS RFC service areas and the NDGD domain. Presently, the QPF products are issued twice daily from 0000 and 1200 GMT for 6-h projections in the range 12 - 120 hours. Plans call for expansion of the temporal coverage to the four six-hourly NCEP cycles and to projections up to 192 hours.
Precip Amt Prob
The basic predicted element is the probability of occurrence of a 6-h threshold precipitation amount category. Eight thresholds, consisting of ≥0.01, ≥0.10, ≥0.25, ≥0.50, ≥0.75, ≥1.00, ≥1.50, and ≥2.00 inches, are predicted. A probability graphic for each threshold and each forecast projection is provided.
Derived Precip Amt
Three precipitation amount variables are derived from the (basic) probabilities for threshold precipitation amounts: precipitation amount category, precipitation amount, and probability-weighted precipitation amount. A graphic for each product and each projection is provided.
The precipitation amount category is a coded precipitation variable, which denotes an interval/category of forecast precipitation amount. It is derived from the probabilities for precipitation amount thresholds through two steps. (1) A pre-specified threshold probability is applied to the probability for the corresponding precipitation threshold. (The threshold probability for a category is computed from historical data whereby the threat score is optimized within a bias range of 0.95 - 1.25.) This is done for each precipitation threshold to determine which (overlapping) categories may occur. (2) The precipitation amount category is specified via a decision tree algorithm to determine the heaviest category that is likely to occur. The category codes are linked to the precipitation amount as:
The precipitation amount is a continuous precipitation variable that consists of interpolated precipitation amounts between the predicted (bounded) categories and extrapolated amounts above the top threshold (unbounded) category. This variable is computed from a prescribed algebraic formula, wherein the dominant variable is the predicted precipitation amount category; secondary variables include the associated probability and threshold probability. This precipitation amount variable is effective in specifying the heaviest precipitation in an area, but its weakness is that it can under-forecast areas of very light precipitation or precipitation that may occur where the probabilities for all categories fall below the associated thresholds (occurs most often in a convective environment where the precipitation threat is weak).
The probability-weighted precipitation amount is computed as the sum of products of the probabilities for discrete precipitation intervals and conditional climatic mean precipitation amounts within these intervals; the summation is over all discrete intervals. This probability-weighted precipitation variable is effective in delineating broad areas of light or scattered convective precipitation, but it greatly under-forecasts heavy amounts.
It is noted that these three precipitation variables complement one another in two fundamental ways. (1) The precipitation amount category should be useful for interpreting the precipitation amount because the latter is derived from the former. (2) The strength of the precipitation amount is in specifying the coverage of heavy precipitation amounts, whereas the strength of probability-weighted precipitation amount is in delineating the areal coverage where any type of precipitation (light or heavy) is a threat.